From: Chih-Jen Lin <cjlin@csie.ntu.edu.tw>
Date: Wed, 28 Oct 2020 09:01:51 +0000 (+0800)
Subject: Stopping conditon of dual CD method for L1- and L2-loss SVM is changed
X-Git-Tag: v242~1
X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=475dfdbae1fa61d6507e9e5c9d5e132b52e00de5;p=liblinear

Stopping conditon of dual CD method for L1- and L2-loss SVM is changed

Currently the stopping condition is M-m <= \epsilon, where

M = max proj_gard(alpha) and m = min proj_grad(alpha).

This condition works well in practice, but we must ensure that the
following situation does not occur.

  M, m >> 0  or M,m << 0
but
  M - m <= \epsilon.

From the optimality condition, |M| and |m| must be close to zero.
For example, because

  grad_i f(0) = -1, for all i.

M=m= -1 and they satisfy M-m <= \epsilon.  Thus we add |M| <=\epsilon
and |m| <= \epsilon$ into the stopping condition.  In fact we can use
only these two inequalities as the stopping condition, but for
historical reasons we keep M-m <= \epsilon as the main condition to
check.
---

diff --git a/linear.cpp b/linear.cpp
index 86ec3dd..5426533 100644
--- a/linear.cpp
+++ b/linear.cpp
@@ -1025,7 +1025,8 @@ static int solve_l2r_l1l2_svc(const problem *prob, const parameter *param, doubl
 		if(iter % 10 == 0)
 			info(".");
 
-		if(PGmax_new - PGmin_new <= eps)
+		if(PGmax_new - PGmin_new <= eps &&
+			fabs(PGmax_new) <= eps && fabs(PGmin_new) <= eps)
 		{
 			if(active_size == l)
 				break;